Audiobus: Use your music apps together.

What is Audiobus?Audiobus is an award-winning music app for iPhone and iPad which lets you use your other music apps together. Chain effects on your favourite synth, run the output of apps or Audio Units into an app like GarageBand or Loopy, or select a different audio interface output for each app. Route MIDI between apps — drive a synth from a MIDI sequencer, or add an arpeggiator to your MIDI keyboard — or sync with your external MIDI gear. And control your entire setup from a MIDI controller.

Download on the App Store

Audiobus is the app that makes the rest of your setup better.

Interesting old article entitled "Of Filters & Phase Relationships" from 1999

Link:
https://www.soundonsound.com/techniques/filters-phase-relationships

This article is an examination of analog synth filters.

My understanding of it, is that old analog synth filters introduce a degree of phase shift into the waveform passing through the filter, and that degree of phase shift changes in respect to the frequency bands of the waveform passing through the filter.

The result is a filter that is in a continuous state of actively phase shifting frequency bands in the entire signal, at different degrees of shift, relative to the frequency bands passing through it.

As one might imagine, this could result in a very complex frequency spectrum based interactivity between differing phase relationships.

This may help explain why the old analog synth filters provided a "warmer" filter response compared to the modern digital variety.

Of course, now I need to search for ways to emulate this type of filter response using drambo and/or other iPadOS tools.

What's needed is a way of dividing up and identifying the bandwidths within an audio signal, then applying degrees of phase shift to only selected frequency bands. The more control over the phase shift response, the better.

Anyone have any ideas?

The Drambo Delay module can introduce phase shift. But how to achieve active "relative" band following frequency phase shift, that's not just a frequency based array of "stepped" static phase settings?

I think the ideal would be continuously variable phasing degrees spread throughout the entire filtered signal spectrum.

How does one do that?

Comments

  • edited January 2021

    Not sure you need to emulate anything... as the signal processing codes including all the filter types are the basic building blocks of audio apps such as Drambo, e.g., if you use a LPF, both the frequency and phase responses are “baked-in”, so mixing filters will get you the complex interactions as described in the article. In fact, DSP can be used to create filters that can not exist in the analog world, perhaps that’s what you are after?

  • I'm no dsp expert, but maybe all pass filters? A static APF will just selectively shift the phase of select frequencies. Modulating an APF (with an envelope or waveform) creates a subtle detuned sound. Otherwise the 'effect' is almost imperceptible. If you were to put several APF in parallel, each fed with a random generator, or S&H, you may be able to get slightly unpredictable phase relationships. I don't know if this is what you're after, but phase is the domain of APF, so I figure there must be something there to discover.

  • edited January 2021

    Not sure if this is what the OP wants... but For the most “potent” filter with very complex frequency and phase responses, check out the Waveguide in Drambo.

    Waveguide is a karplus strong on steroids... it adds allpass filter and optional nonlinearity.. waveguide model is extended. Its good for drums.. and inharmonic stuff.. play with apf , filter and modulate freq / filter.

  • @horsetrainer said:
    Link:
    https://www.soundonsound.com/techniques/filters-phase-relationships

    This article is an examination of analog synth filters.

    My understanding of it, is that old analog synth filters introduce a degree of phase shift into the waveform passing through the filter, and that degree of phase shift changes in respect to the frequency bands of the waveform passing through the filter.

    Digital filters do too. There is a way to not have phase shift, and sometimes you want that. There's not really a difference between analog and digital here.

    This may help explain why the old analog synth filters provided a "warmer" filter response compared to the modern digital variety.

    No, this is more due analog filters (and digital filters which emulate this) being non-linear and having distortion (including on the feedback path) that a simple digital filter doesn't have. It can be emulated in digital, but it requires either writing code or something like Audulus. It also requires at least some math. I don't think you could do that in Drambo - it's not quite low level enough.

  • edited January 2021
    The user and all related content has been deleted.
  • edited January 2021
    The user and all related content has been deleted.
  • The user and all related content has been deleted.
  • The part of the article that interests me. Is the general concept that analog filter design, might shift the phase relationships between harmonics to degrees that fall outside of what one would calculate using "perfect" filter math.

    For example, in a case where math would predict the relationship between two harmonics within the overall signal to be 180° out of phase, they would actually measure ~170°.

    Then consider the multitude of harmonic relationships within a given complex signal, and imagine some can be left "pure", while other relationships can have those phase relationships deliberately skewed out of phase, by some adjustable amount.

    That's the basic premise of the general type of experimentation I'm interested in doing.... Testing the effects of altering phase between harmonics on a selective frequency basis.

    But I'm seeing two ways to go about this:

    1) Send a signal through a series of parallel band pass filters, and alter the phase of the output of each by some degree, before sending all outputs to a mixer.

    2) What I'm really interested in.... Coming up with a method to use the "amplitude" of a given frequency, to modulate a degree of phase shift applied to either that frequency, or any other frequency within the given signal.

    Basically I'd like to find a method for isolating variable widths of band passed frequencies, and converting the amplitude of those isolated variable widths of frequency into a Control Voltage, of a corresponding (relative) amplitude.

    Then I can use that CV as a modulation source for use for all kind of experiments.

  • Wasn't that also the title of an OMD album?

  • I'm running multiple BP filters in Drambo into a Full Rectify and then a Slew Limiter to convert signal amplitude into CV.
    The CV then controls Phase Shift using a Delay Module.
    The original signal is then mixed with the phase shifted signal, and into a final LP filter.

    The results are very good. I'm already getting sudo OBX-like low end filter response. The 1999 article seems to provide some useful info for vintage filter emulation.

  • @horsetrainer said:
    The part of the article that interests me. Is the general concept that analog filter design, might shift the phase relationships between harmonics to degrees that fall outside of what one would calculate using "perfect" filter math.

    It‘s cool if an article inspires you to new ideas, but this particular essay just describes the most basic aspects of theory.
    Not a single design is even mentioned.
    Regarding perfect filter math there’s a quote (one of my favourites) from Christoph Kemper, designer of the Access Virus (one of the first virtual analog synths available):
    it’s easy to pick a math book and program a perfect algorithm - but it takes experience to know which sounds convincing, too... and those are often not the mathematically most correct ones.
    The SoS article leaves this as an exercise to the reader ;)

  • edited January 2021

    I want to throw the app bandshift into the room. you have around 30 bands and can assign each band an individual lfo to modulate pitch, lfo rate can also be zero. and you can automate each parameter for each band. I guess it's not exactly what you want but idk, maybe this app combined with a filter might give similar results to what you are looking for...

  • edited January 2021

    @Telefunky said:

    @horsetrainer said:
    The part of the article that interests me. Is the general concept that analog filter design, might shift the phase relationships between harmonics to degrees that fall outside of what one would calculate using "perfect" filter math.

    It‘s cool if an article inspires you to new ideas, but this particular essay just describes the most basic aspects of theory.
    Not a single design is even mentioned.
    Regarding perfect filter math there’s a quote (one of my favourites) from Christoph Kemper, designer of the Access Virus (one of the first virtual analog synths available):
    it’s easy to pick a math book and program a perfect algorithm - but it takes experience to know which sounds convincing, too... and those are often not the mathematically most correct ones.
    The SoS article leaves this as an exercise to the reader ;)

    I'm definitely looking at that article through a lens of my own experiences of tinkering with analog circuits when I was younger. The article did show a basic two-component passive LP resistor/capacitor circuit.

    That instantly made me remember building simple transistor 180° RC phase shift feedback oscillators many years ago. So you're right, the article was mostly a source of inspiration to investigate phase shift just to see what it might sound like.

    But I think the hypothesis is worthy of more study. The primary reason is because I think it's reasonable to consider that the analog circuit designs used in vintage synths, just might have caused delay to be introduced by the inherent response characteristics of the components used in that era.

    I want to do further study into this. But I can say from my first experiments that the effect of inducing phase delay into sequestered frequency regions is capable of producing some surprisingly powerful signal alterations.... And it don't sound too bad either.... :smile:

    If no developer has yet produced an effect based on these principles, I think some might want to look into it.

    Far more could be done than just phase shifting isolated frequency bands using selected frequency signal amplitude as a modulator. Phase shifting could also be modulated on a basis of selectively converting pitch to CV as well.

    I think this selective frequency phase shifting hypothesis might also be understood from a perspective of general waveform theory. Where it's said that any wave-shape can be reproduced by combining a proper organization and number of sign waves.

    What interests me the most, is that using selected aspects of an entire audio signal to control signal modification. Is a methodology based in a form of feedback. The characteristics of the sound entering the "filter" determine the modification of the sound.

    A filter could be designed to have programable x/y knob matrix, where select x-input frequency(s) could be routed to apply a relative degree of phase shift to select y-output frequency(s). It's almost similar to FM, but not quite, because it's working with existing sounds to modify them, more like a type of effect.

  • @dobbs said:
    I want to throw the app bandshift into the room. you have around 30 bands and can assign each band an individual lfo to modulate pitch, lfo rate can also be zero. and you can automate each parameter for each band. I guess it's not exactly what you want but idk, maybe this app combined with a filter might give similar results to what you are looking for...

    Thanks for reminding me about the virsyn banddelay and bandshift apps.

    I was looking at those, but did not notice the potential of phase shifting by using selected bands of the input signal as the methodology of controlling the degree of shift applied to selected output bands.

  • edited January 2021
    The user and all related content has been deleted.
  • edited January 2021

    @horsetrainer said:
    I was looking at those, but did not notice the potential of phase shifting by using selected bands of the input signal as the methodology of controlling the degree of shift applied to selected output bands.

    Just forget about this phase shift terminolgy - it points into the wrong direction... and it applies only to cyclic signals anyway.
    In the end it‘s all about very short delays, a shift in timing.
    Your observation about it‘s effects are correct - it‘s a crucial part of digital audio processing.

    Multi-channel streams should remain sample accurate throughout processing, which is not as easy as it may read.
    High end reverb processors include the effects of changing air density that constantly „modulates“ the effective runtime of signals in real life by sample delays.
    If you have an acoustic guitar recorded with 2 microphones and you displace the tracks (equivalent to shifting mics) for just a couple of samples, the instrument may sound very different. etc...

  • edited January 2021
    The user and all related content has been deleted.
  • edited January 2021

    Of course it‘s unpredictable before you shift one of the tracks, but you immediately hear the result (and the process can be reversed anyway).
    But it‘s not playing with phase cancelling, it‘s equally playing with phase emphasizing... and everything between... digital eq-ing with many bands ;)
    (as every frequency has it‘s very own specific time-factor - they once used to write micro-seconds instead of Hz in circuit diagrams)

    it‘s not a suggestion to ignore proper mic positioning, but an example that‘s easy to test how much influence 3, 5, 10, ... samples may have.

Sign In or Register to comment.